Replace with repaired direct-answer SFT plus latest-SFT DWM alpha 0.1

README.md

@@ -16,7 +16,7 @@ model_name: Qwen3.8-27B-SwissNeuron-Derisked
 
 # Qwen3.8-27B-SwissNeuron-Derisked
 
-**Qwen3.8-27B-SwissNeuron-Derisked**, known as **SwissNeuron**, is a BF16, 27B-parameter Qwen3.8 derivative engineered in Switzerland for direct technical work, strong reasoning, and capability retention. It combines focused in-house post-training with a conservative internal geometric derisk procedure, then extends the model configuration to a 1,048,576-token context window using factor-4 YaRN/RoPE scaling.
+**Qwen3.8-27B-SwissNeuron-Derisked**, known as **SwissNeuron**, is a BF16, 27B-parameter Qwen3.8 derivative engineered in Switzerland for direct technical work, strong reasoning, and capability retention. This revision uses a repaired direct-answer SFT corpus with the earlier judge/verifier register removed, followed by a conservative internal geometric derisk procedure. The model configuration provides a 1,048,576-token context window using factor-4 YaRN/RoPE scaling.
 
 SwissNeuron is intended to provide Swiss-quality model engineering: precise provenance, conservative weight surgery, reproducible artifacts, and transparent limitations.
 
@@ -38,9 +38,9 @@ This release should be served with **thinking disabled** for ordinary chat, codi
 enable_thinking=False
 ```
 
-The focused post-training corpus contains a high concentration of rethink and answer-verification trajectories. In the current BF16 release, enabling the native thinking channel can therefore make simple requests produce an overly rigid internal verification trace (for example, `Step 1`, `EXPECTED VERIFICATION`, and `ORIGINAL ANSWER`) before the otherwise normal final answer. A system prompt does not reliably suppress this learned thinking-mode format. Non-thinking mode does not exhibit the issue in our direct probes and returns the concise final answer normally.
+The repaired post-training corpus removes the earlier judge/verifier envelope from the model-facing answer targets. In direct probes of this revision, native thinking produces ordinary concise reasoning for simple greetings and coding requests rather than an `EXPECTED VERIFICATION` / `ORIGINAL ANSWER` grading trace. Non-thinking mode remains supported for the lowest-latency agent integrations.
 
-For agent harnesses, set `enable_thinking=False` in the Qwen chat-template arguments rather than stripping the trace after generation. The planned larger fine-tune will rebalance native reasoning, rethink, and direct-answer data before advertising thinking-mode agent compatibility.
+For maximum compatibility with agent harnesses, `enable_thinking=False` remains a valid deployment choice. Preserve the included chat template rather than stripping generated reasoning after the fact.
 
 ## What makes SwissNeuron different
 
@@ -154,7 +154,7 @@ The larger fine-tuned release will be published as a separate checkpoint rather
 ## Limitations
 
 - The 1M window is enabled through YaRN configuration; extreme-context quality has not yet been comprehensively validated.
-- Native thinking mode can overproduce a rigid verification/critic trace because of the rethink-heavy post-training mixture. Use `enable_thinking=False` for general chat and agent integrations in this release.
+- Native thinking was regression-tested on simple greeting and coding prompts after the corpus repair; users should still validate their serving stack and prompt format.
 - This is a full BF16 release and requires substantial accelerator memory.
 - Outputs may be inaccurate, incomplete, or unsuitable for a particular domain. Independently verify consequential results.
 - The repository is public, but users remain responsible for evaluating suitability, licensing requirements, and deployment risk in their own environment.

args.json

@@ -1,5 +1,5 @@
 {
-  "output_dir": "/workspace/train-qwen38/models/karma-qwen38-nonrefusal-sft/v0-20260823-131002",
+  "output_dir": "/workspace/train-qwen38/models/direct-repair-sft/v0-20260823-225101",
   "per_device_train_batch_size": 1,
   "num_train_epochs": 1.0,
   "max_steps": -1,
@@ -46,7 +46,7 @@
   "report_to": [
     "tensorboard"
   ],
-  "run_name": "/workspace/train-qwen38/models/karma-qwen38-nonrefusal-sft/v0-20260823-131002",
+  "run_name": "/workspace/train-qwen38/models/direct-repair-sft/v0-20260823-225101",
   "project": "huggingface",
   "trackio_space_id": null,
   "trackio_bucket_id": null,
@@ -153,7 +153,7 @@
   "do_predict": false,
   "resume_from_checkpoint": null,
   "warmup_ratio": null,
-  "logging_dir": "/workspace/train-qwen38/models/karma-qwen38-nonrefusal-sft/v0-20260823-131002/runs",
+  "logging_dir": "/workspace/train-qwen38/models/direct-repair-sft/v0-20260823-225101/runs",
   "local_rank": 0,
   "sortish_sampler": false,
   "predict_with_generate": false,
@@ -211,7 +211,7 @@
   "use_ray": false,
   "ray_exp_name": null,
   "device_groups": null,
-  "model": "/root/Qwen3.8-27B",
+  "model": "/workspace/train-qwen38/models/Qwen3.8-27B",
   "model_type": "qwen3_5",
   "model_revision": null,
   "task_type": "causal_lm",
@@ -393,12 +393,12 @@
   "global_world_size": 8,
   "local_world_size": 8,
   "model_suffix": "Qwen3.8-27B",
-  "model_info": "ModelInfo(model_type='qwen3_5', model_dir='/root/Qwen3.8-27B', torch_dtype=torch.bfloat16, max_model_len=262144, quant_method=None, quant_bits=None, rope_scaling=None, is_moe_model=False, is_multimodal=True, config=None, task_type='causal_lm', num_labels=None)",
+  "model_info": "ModelInfo(model_type='qwen3_5', model_dir='/workspace/train-qwen38/models/Qwen3.8-27B', torch_dtype=torch.bfloat16, max_model_len=262144, quant_method=None, quant_bits=None, rope_scaling=None, is_moe_model=False, is_multimodal=True, config=None, task_type='causal_lm', num_labels=None)",
   "model_meta": "ModelMeta(model_type='qwen3_5', model_groups=[ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3.5-0.8B', hf_model_id='Qwen/Qwen3.5-0.8B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-2B', hf_model_id='Qwen/Qwen3.5-2B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-4B', hf_model_id='Qwen/Qwen3.5-4B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-9B', hf_model_id='Qwen/Qwen3.5-9B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-27B', hf_model_id='Qwen/Qwen3.5-27B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-27B-FP8', hf_model_id='Qwen/Qwen3.5-27B-FP8', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-0.8B-Base', hf_model_id='Qwen/Qwen3.5-0.8B-Base', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-2B-Base', hf_model_id='Qwen/Qwen3.5-2B-Base', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-4B-Base', hf_model_id='Qwen/Qwen3.5-4B-Base', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.5-9B-Base', hf_model_id='Qwen/Qwen3.5-9B-Base', model_path=None, ms_revision=None, hf_revision=None)], template='qwen3_5', ignore_patterns=None, requires=None, tags=[]), ModelGroup(models=[Model(ms_model_id='Qwen/Qwen3.6-27B', hf_model_id='Qwen/Qwen3.6-27B', model_path=None, ms_revision=None, hf_revision=None), Model(ms_model_id='Qwen/Qwen3.6-27B-FP8', hf_model_id='Qwen/Qwen3.6-27B-FP8', model_path=None, ms_revision=None, hf_revision=None)], template='qwen3_5', ignore_patterns=None, requires=None, tags=[])], loader=<class 'swift.model.models.qwen.Qwen3_5Loader'>, template=None, model_arch=MultiModelKeys(arch_name='qwen2_vl', embedding=None, module_list=None, lm_head=None, q_proj=None, k_proj=None, v_proj=None, o_proj=None, attention=None, mlp=None, down_proj=None, qkv_proj=None, qk_proj=None, qa_proj=None, qb_proj=None, kv_proj=None, kva_proj=None, kvb_proj=None, language_model=['model.language_model', 'lm_head'], aligner=['model.visual.merger'], vision_tower=['model.visual'], generator=[]), mcore_model_type=None, architectures=['Qwen3_5ForConditionalGeneration'], additional_saved_files=[], torch_dtype=None, is_multimodal=True, is_reward=False, task_type=None, ignore_patterns=None, requires=['transformers>=5.0.0.dev', 'qwen_vl_utils>=0.0.14', 'decord'], tags=['vision', 'video'])",
-  "model_dir": "/root/Qwen3.8-27B",
+  "model_dir": "/workspace/train-qwen38/models/Qwen3.8-27B",
   "template_meta": "QwenTemplateMeta(template_type='qwen3_5', prefix=[], prompt=['<|im_start|>user\\n{{QUERY}}<|im_end|>\\n<|im_start|>assistant\\n'], chat_sep=['<|im_end|>\\n'], suffix=['<|im_end|>\\n'], template_cls=<class 'swift.template.templates.qwen.Qwen3_5Template'>, system_prefix=['<|im_start|>system\\n{{SYSTEM}}<|im_end|>\\n'], default_system=None, auto_add_bos=False, stop_words=['<|endoftext|>'], agent_template='qwen3_5', is_thinking=True, thinking_prefix='<think>\\n', non_thinking_prefix='<think>\\n\\n</think>\\n\\n', history_thinking_prefix='')",
   "_val_dataset_exists": true,
   "hub": "<class 'swift.hub.hub.HFHub'>",
   "evaluation_strategy": "steps",
-  "training_args": "Seq2SeqTrainingArguments(output_dir='/workspace/train-qwen38/models/karma-qwen38-nonrefusal-sft/v0-20260823-131002', per_device_train_batch_size=1, num_train_epochs=1.0, max_steps=-1, learning_rate=0.0001, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_steps=5.0, optim=<OptimizerNames.ADAMW_TORCH_FUSED: 'adamw_torch_fused'>, optim_args=None, weight_decay=0.01, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, optim_target_modules=None, gradient_accumulation_steps=2, average_tokens_across_devices=None, max_grad_norm=1.0, label_smoothing_factor=0.0, bf16=True, fp16=False, bf16_full_eval=False, fp16_full_eval=False, tf32=None, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, use_liger_kernel=True, liger_kernel_config=None, use_cache=False, neftune_noise_alpha=None, torch_empty_cache_steps=None, auto_find_batch_size=False, logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_steps=5, logging_first_step=True, log_on_each_node=True, logging_nan_inf_filter=True, include_num_input_tokens_seen=None, log_level='passive', log_level_replica='warning', disable_tqdm=False, report_to=['tensorboard'], run_name='/workspace/train-qwen38/models/karma-qwen38-nonrefusal-sft/v0-20260823-131002', project='huggingface', trackio_space_id=None, trackio_bucket_id=None, trackio_static_space_id=None, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, eval_steps=50, eval_delay=0, per_device_eval_batch_size=1, prediction_loss_only=False, eval_on_start=False, eval_do_concat_batches=True, eval_use_gather_object=False, eval_accumulation_steps=None, include_for_metrics=[], batch_eval_metrics=False, save_only_model=False, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=50, save_on_each_node=False, save_total_limit=8, enable_jit_checkpoint=False, push_to_hub=False, hub_token=None, hub_private_repo=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_always_push=False, hub_revision=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, restore_callback_states_from_checkpoint=False, full_determinism=False, seed=42, data_seed=42, use_cpu=False, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), parallelism_config=None, dataloader_drop_last=False, dataloader_num_workers=4, dataloader_pin_memory=True, dataloader_persistent_workers=False, dataloader_prefetch_factor=2, remove_unused_columns=False, label_names=None, train_sampling_strategy='random', length_column_name='length', ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, ddp_static_graph=None, ddp_backend=None, ddp_timeout=18000000, fsdp=[], fsdp_config=None, deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'offload_param': {'device': 'none', 'pin_memory': True}, 'overlap_comm': False, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'zero_quantized_weights': False, 'zero_quantized_gradients': False, 'stage3_prefetch_bucket_size': 'auto', 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, debug=[], skip_memory_metrics=True, do_train=False, do_eval=True, do_predict=False, resume_from_checkpoint=None, warmup_ratio=None, logging_dir='/workspace/train-qwen38/models/karma-qwen38-nonrefusal-sft/v0-20260823-131002/runs', local_rank=0, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, tuner_backend='peft', vit_gradient_checkpointing=False, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, safe_serialization=True, max_shard_size='5GB', check_model=False, acc_strategy='token', train_dataloader_shuffle=True, group_by_length=False, max_epochs=None, aligner_lr=None, vit_lr=None, use_logits_to_keep=None, ds3_gather_for_generation=True, resume_only_model=False, optimizer=None, loss_type=None, mrl_dims=None, eval_metric=None, callbacks=[], early_stop_interval=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, tuner_type='lora', use_galore=False, galore_target_modules=None, galore_rank=128, galore_update_proj_gap=50, galore_scale=1.0, galore_proj_type='std', galore_optim_per_parameter=False, galore_with_embedding=False, galore_quantization=False, galore_proj_quant=False, galore_proj_bits=4, galore_proj_group_size=256, galore_cos_threshold=0.4, galore_gamma_proj=2, galore_queue_size=5, lisa_activated_layers=0, lisa_step_interval=20, use_flash_ckpt=False)"
+  "training_args": "Seq2SeqTrainingArguments(output_dir='/workspace/train-qwen38/models/direct-repair-sft/v0-20260823-225101', per_device_train_batch_size=1, num_train_epochs=1.0, max_steps=-1, learning_rate=0.0001, lr_scheduler_type=<SchedulerType.COSINE: 'cosine'>, lr_scheduler_kwargs=None, warmup_steps=5.0, optim=<OptimizerNames.ADAMW_TORCH_FUSED: 'adamw_torch_fused'>, optim_args=None, weight_decay=0.01, adam_beta1=0.9, adam_beta2=0.95, adam_epsilon=1e-08, optim_target_modules=None, gradient_accumulation_steps=2, average_tokens_across_devices=None, max_grad_norm=1.0, label_smoothing_factor=0.0, bf16=True, fp16=False, bf16_full_eval=False, fp16_full_eval=False, tf32=None, gradient_checkpointing=True, gradient_checkpointing_kwargs=None, torch_compile=False, torch_compile_backend=None, torch_compile_mode=None, use_liger_kernel=True, liger_kernel_config=None, use_cache=False, neftune_noise_alpha=None, torch_empty_cache_steps=None, auto_find_batch_size=False, logging_strategy=<IntervalStrategy.STEPS: 'steps'>, logging_steps=5, logging_first_step=True, log_on_each_node=True, logging_nan_inf_filter=True, include_num_input_tokens_seen=None, log_level='passive', log_level_replica='warning', disable_tqdm=False, report_to=['tensorboard'], run_name='/workspace/train-qwen38/models/direct-repair-sft/v0-20260823-225101', project='huggingface', trackio_space_id=None, trackio_bucket_id=None, trackio_static_space_id=None, eval_strategy=<IntervalStrategy.STEPS: 'steps'>, eval_steps=50, eval_delay=0, per_device_eval_batch_size=1, prediction_loss_only=False, eval_on_start=False, eval_do_concat_batches=True, eval_use_gather_object=False, eval_accumulation_steps=None, include_for_metrics=[], batch_eval_metrics=False, save_only_model=False, save_strategy=<SaveStrategy.STEPS: 'steps'>, save_steps=50, save_on_each_node=False, save_total_limit=8, enable_jit_checkpoint=False, push_to_hub=False, hub_token=None, hub_private_repo=None, hub_model_id=None, hub_strategy=<HubStrategy.EVERY_SAVE: 'every_save'>, hub_always_push=False, hub_revision=None, load_best_model_at_end=False, metric_for_best_model='loss', greater_is_better=False, ignore_data_skip=False, restore_callback_states_from_checkpoint=False, full_determinism=False, seed=42, data_seed=42, use_cpu=False, accelerator_config=AcceleratorConfig(split_batches=False, dispatch_batches=False, even_batches=True, use_seedable_sampler=True, non_blocking=False, gradient_accumulation_kwargs=None, use_configured_state=False), parallelism_config=None, dataloader_drop_last=False, dataloader_num_workers=4, dataloader_pin_memory=True, dataloader_persistent_workers=False, dataloader_prefetch_factor=2, remove_unused_columns=False, label_names=None, train_sampling_strategy='random', length_column_name='length', ddp_find_unused_parameters=None, ddp_bucket_cap_mb=None, ddp_broadcast_buffers=None, ddp_static_graph=None, ddp_backend=None, ddp_timeout=18000000, fsdp=[], fsdp_config=None, deepspeed={'fp16': {'enabled': 'auto', 'loss_scale': 0, 'loss_scale_window': 1000, 'initial_scale_power': 16, 'hysteresis': 2, 'min_loss_scale': 1}, 'bf16': {'enabled': 'auto'}, 'zero_optimization': {'stage': 3, 'offload_optimizer': {'device': 'none', 'pin_memory': True}, 'offload_param': {'device': 'none', 'pin_memory': True}, 'overlap_comm': False, 'contiguous_gradients': True, 'sub_group_size': 1000000000.0, 'reduce_bucket_size': 'auto', 'zero_quantized_weights': False, 'zero_quantized_gradients': False, 'stage3_prefetch_bucket_size': 'auto', 'stage3_param_persistence_threshold': 'auto', 'stage3_max_live_parameters': 1000000000.0, 'stage3_max_reuse_distance': 1000000000.0, 'stage3_gather_16bit_weights_on_model_save': True}, 'gradient_accumulation_steps': 'auto', 'gradient_clipping': 'auto', 'steps_per_print': 2000, 'train_batch_size': 'auto', 'train_micro_batch_size_per_gpu': 'auto', 'wall_clock_breakdown': False}, debug=[], skip_memory_metrics=True, do_train=False, do_eval=True, do_predict=False, resume_from_checkpoint=None, warmup_ratio=None, logging_dir='/workspace/train-qwen38/models/direct-repair-sft/v0-20260823-225101/runs', local_rank=0, sortish_sampler=False, predict_with_generate=False, generation_max_length=None, generation_num_beams=None, generation_config=None, tuner_backend='peft', vit_gradient_checkpointing=False, router_aux_loss_coef=0.0, enable_dft_loss=False, enable_channel_loss=False, safe_serialization=True, max_shard_size='5GB', check_model=False, acc_strategy='token', train_dataloader_shuffle=True, group_by_length=False, max_epochs=None, aligner_lr=None, vit_lr=None, use_logits_to_keep=None, ds3_gather_for_generation=True, resume_only_model=False, optimizer=None, loss_type=None, mrl_dims=None, eval_metric=None, callbacks=[], early_stop_interval=None, eval_use_evalscope=False, eval_dataset=[], eval_dataset_args=None, eval_limit=None, eval_generation_config=None, extra_eval_args=None, tuner_type='lora', use_galore=False, galore_target_modules=None, galore_rank=128, galore_update_proj_gap=50, galore_scale=1.0, galore_proj_type='std', galore_optim_per_parameter=False, galore_with_embedding=False, galore_quantization=False, galore_proj_quant=False, galore_proj_bits=4, galore_proj_group_size=256, galore_cos_threshold=0.4, galore_gamma_proj=2, galore_queue_size=5, lisa_activated_layers=0, lisa_step_interval=20, use_flash_ckpt=False)"
 }

dwm_provenance.json

@@ -1,33 +1,33 @@
 {
   "alpha": 0.1,
   "audit": {
-    "component_ratio_max": 0.9202628135681152,
-    "component_ratio_mean": 0.9149049116719153,
-    "component_ratio_min": 0.9011346697807312,
-    "norm_ratio_max": 1.0000070333480835,
-    "norm_ratio_mean": 0.9999727582739245,
-    "norm_ratio_min": 0.9998393058776855,
-    "relative_delta_max": 0.006776769645512104,
-    "relative_delta_mean": 0.0019126400884781633,
-    "relative_delta_min": 0.0013704006560146809,
-    "signed_retention_max": 0.9201236963272095,
-    "signed_retention_mean": 0.9147886436793112,
-    "signed_retention_min": 0.9011276960372925
+    "component_ratio_max": 0.9210951328277588,
+    "component_ratio_mean": 0.9151080443013099,
+    "component_ratio_min": 0.9009799957275391,
+    "norm_ratio_max": 1.0000025033950806,
+    "norm_ratio_mean": 0.9999723818994337,
+    "norm_ratio_min": 0.999819815158844,
+    "relative_delta_max": 0.0075966790318489075,
+    "relative_delta_mean": 0.0019452353552614728,
+    "relative_delta_min": 0.0013362898025661707,
+    "signed_retention_max": 0.9209601283073425,
+    "signed_retention_mean": 0.9149930731904122,
+    "signed_retention_min": 0.9009749293327332
   },
   "checkpoint_contains_recipe_metadata": false,
   "direction_run_id": null,
-  "direction_sha256": "c754628ca259d9cfe55af81f93788f403424782f05994c5940ebab7613d81724",
-  "directions": "/workspace/train-qwen38/derisk/bank_final_nonrefusal.pt",
+  "direction_sha256": "6e2f8a5bed1856b7ced8f47ed27647a433b30435838632a2fdea595690944e7b",
+  "directions": "/workspace/train-qwen38/derisk/bank_direct_repair_sft.pt",
   "edited_matrices": 124,
   "hidden_size": 5120,
   "layers": 64,
   "method": "soft directional weight modification",
   "nominal_directional_retention": 0.9,
-  "output": "/workspace/train-qwen38/models/karma-qwen38-nonrefusal-derisk-a01",
+  "output": "/workspace/train-qwen38/models/direct-repair-sft-derisk-a01",
   "passes": 1,
   "secondary": null,
   "skip_early": 2,
-  "source": "/workspace/train-qwen38/models/karma-qwen38-nonrefusal-sft-SNAPSHOT",
+  "source": "/workspace/train-qwen38/models/direct-repair-sft-merged",
   "source_index_sha256": "a600c6626eb1eb653a3f078ab3dd1c7a8fab2e71185c7ef3aa923a39aaa637be",
   "source_modified": false,
   "surface_counts": {

internal_direction_bank.pt

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c754628ca259d9cfe55af81f93788f403424782f05994c5940ebab7613d81724
-size 1313045
+oid sha256:6e2f8a5bed1856b7ced8f47ed27647a433b30435838632a2fdea595690944e7b
+size 1313117

model-00002-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f9d51464f2006c4a12240f251e22a55e894cf0db35be8f5fbd5295340ea8dee2
+oid sha256:897975c55d40ad2bfb27555c9fb43aa59ed67ad037e7d930457e051a50f2d450
 size 4842451920

model-00003-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f84e5d7a394117f4b98354754b2edaa3a88fc0f495db7832a3f03d3014c2c7ab
+oid sha256:3e1e82a608a4eb770b73b08f20a520ba7f3386353129c151ebdcc35d2a2c79f1
 size 4965227944

model-00004-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:69330da8367eedd733d414aac86af1ab0c540f7f0ba1a9cf8587e4ad93148056
+oid sha256:7a4e031a16791b774d6026e732d934338147ec619dde1e3edd4ea591e5437405
 size 4912819264

model-00005-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:8d5861388eb9f3507e6abdb83d2b72648a27642978e4620038f8724c05cdfdc6
+oid sha256:baf6bf0914f2c60c1ee8009acf3a7798aacc48c23d398907585692790da0255c
 size 4986198544

model-00006-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:777c23c5a7b091eb3bdfaa2cd91d582b998e33ef4f7301dee36ea1da301d0e03
+oid sha256:d344daae69c6bb5b0ce881b677d6834ed36364db0cb2c236738420dcf59b24e8
 size 4912819320

model-00007-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:bff0dd48aeaf81f13c872a799ce8443f2d47cb16c8948730ff8b466b8b2888be
+oid sha256:4a227f9c5ab68adb3485f1fc7612728f9dda8a79d6acb1d17beea0689f615421
 size 4932703272

model-00008-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:9e2ceae6c13145833cb525c7803c69ccff574abe083378dce0d33849a961e1f3
+oid sha256:674e4a309b74b3a3791ef5ed50322046931b5afcfb396cacff4d34a2de1d7729
 size 4966314576

model-00009-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c3f165523deff743af0ca0a57d4d45cdc2cb9744ff509d1616c62d76d3ad054f
+oid sha256:33c458167ceddf0d20d23b5d5c042d20eea792f22907c78563b7580767c1e4a2
 size 4964162248

model-00010-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:f839b2e0b92cc570ecbab95083398de5b8342c0c9e7756070d91cf5664d6b591
+oid sha256:85a30c492afe74e0653a58b7239e76bb875ea265b40d414657002b04888c1d2e
 size 4933789824

model-00011-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:5ec44a2b3edf78998535dcbe205029ca24ecd7ee3f018e23b369f6c9de75dc8b
+oid sha256:7c01c617a4fbbe77e92bc5ecb3524d719f1170dbd2a11c546bd7c908e6f4dd25
 size 4965228032

model-00012-of-00012.safetensors

@@ -1,3 +1,3 @@
 version https://git-lfs.github.com/spec/v1
-oid sha256:c1d27e7a52aabc93d7d09d24bfb08e3899e1bdbea53282899f461e7293878a49
+oid sha256:63561384127fb2566d36980dba639907b5bd8565d676a82c5b319aa97d3bb843
 size 2789094896